Intuitive Note-taking App for Dyslexic Patients - Image Generation Note <Doodling>

서비스 배포 링크

https://doodling1.web.app/

Github Documentation Link

 https://github.com/doodling-TTI/Doodling

⭐️ Project Overview

Doodling is a web service developed as part of an internship project after receiving deep learning education from the US startup incubator Peoplespace. It aims to assist dyslexic patients in reading and acquiring knowledge more easily by converting notes they write into images for easy storage and viewing. The service utilizes the Fine-Tuned Stable Diffusion model to convert written notes into images.

Service Advertisement Video

Demo Video

👬 Team Composition

• 4 developers: Front-end (1), Back-end (1), Deep learning (2)
• Collaboration Tools: Git, Figma, Notion
• I was responsible for page layout, planning, marketing, data labeling, and model development.

🔨 Responsibilities

I was responsible for the development and planning of the Stable Diffusion model.

• Planning and Marketing Demand Research
• Landing Page Planning and Development
• Stable Diffusion Model Fine-Tuning
• Hugging Face Deployment

📈 Project Planning and Market Research Planning

text2Ilustration_product concept, market research.pdf

🕴️ PitchDeck

Doodling Pitch Deck PPT

⚒️ Technologies and Libraries Used

• Keras
• Pytorch
• Hugging Face

💡 Reflections

• Working with cutting-edge technologies like Chat GPT3.5 and Stable Diffusion allowed me to dream about the future of AI technology. Fine-tuning the Stable Diffusion model using labeled data that we created ourselves was a significant experience, as it allowed us to apply existing models to the project at a relatively low cost.